From 45bfe6ee715c90ee64ffbc90dd048ccbf606d97d Mon Sep 17 00:00:00 2001
From: John Factotum <50942278+johnfactotum@users.noreply.github.com>
Date: Tue, 21 Mar 2023 12:03:39 +0800
Subject: [PATCH] Fix whitespace normalization

- Normalize ASCII whitespace only
- Collapse sequence of *one* or more consecutive whitespace characters
---
 epub.js | 10 +++++++---
 fb2.js  | 10 +++++++---
 2 files changed, 14 insertions(+), 6 deletions(-)

diff --git a/epub.js b/epub.js
index 7df3613..65cda41 100644
--- a/epub.js
+++ b/epub.js
@@ -26,8 +26,12 @@ const MIME = {
 // convert to camel case
 const camel = x => x.toLowerCase().replace(/[-:](.)/g, (_, g) => g.toUpperCase())
 
-// remove leading, trailing, and excess internal whitespace
-const whitespacePreLine = str => str ? str.trim().replace(/\s{2,}/g, ' ') : ''
+// strip and collapse ASCII whitespace
+// https://infra.spec.whatwg.org/#strip-and-collapse-ascii-whitespace
+const normalizeWhitespace = str => str ? str
+    .replace(/[\t\n\f\r ]+/g, ' ')
+    .replace(/^[\t\n\f\r ]+/, '')
+    .replace(/[\t\n\f\r ]+$/, '') : ''
 
 const filterAttribute = (attr, value, isList) => isList
     ? el => el.getAttribute(attr)?.split(/\s/)?.includes(value)
@@ -38,7 +42,7 @@ const filterAttribute = (attr, value, isList) => isList
 const getAttributes = (...xs) => el =>
     el ? Object.fromEntries(xs.map(x => [camel(x), el.getAttribute(x)])) : null
 
-const getElementText = el => whitespacePreLine(el?.textContent)
+const getElementText = el => normalizeWhitespace(el?.textContent)
 
 const childGetter = (doc, ns) => {
     // ignore the namespace if it doesn't appear in document at all
diff --git a/fb2.js b/fb2.js
index f99e8c2..b85277b 100644
--- a/fb2.js
+++ b/fb2.js
@@ -1,5 +1,8 @@
-const trim = str => str?.trim()?.replace(/\s{2,}/g, ' ')
-const getElementText = el => trim(el?.textContent)
+const normalizeWhitespace = str => str ? str
+    .replace(/[\t\n\f\r ]+/g, ' ')
+    .replace(/^[\t\n\f\r ]+/, '')
+    .replace(/[\t\n\f\r ]+$/, '') : ''
+const getElementText = el => normalizeWhitespace(el?.textContent)
 
 const NS = {
     XLINK: 'http://www.w3.org/1999/xlink',
@@ -287,7 +290,8 @@ export const makeFB2 = async blob => {
             const str = template(el.outerHTML)
             const blob = new Blob([str], { type: MIME.XHTML })
             const url = URL.createObjectURL(blob)
-            const title = trim(el.querySelector('.title, .subtitle, p')?.textContent
+            const title = normalizeWhitespace(
+                el.querySelector('.title, .subtitle, p')?.textContent
                 ?? (el.classList.contains('title') ? el.textContent : ''))
             return {
                 ids, title, titles, load: () => url,